This study investigates the impact of bias-correction and downscaling on climate model projections of U.S. maize yields. Using a multi-model ensemble of statistically bias-corrected and downscaled climate models (NEX-GDDP) and their parent CMIP5 models, the researchers drove a statistical panel model of U.S. maize yields. The analysis revealed that CMIP5 models overestimate historical yield variability, while bias-corrected and downscaled versions underestimate the most severe weather-induced yield declines. Significant differences in projected yields and other key metrics throughout the century highlight the trade-offs involved in choosing between different climate model approaches regarding resolution, historical accuracy, and projection confidence.

David C. Lafferty, Ryan L. Sriver, Iman Haqiqi, Thomas W. Hertel, Klaus Keller, Robert E. Nicholas